Claude Sonnet 4.5 vs GPT-5 Nano
Côte à côte. Chaque métrique. Chaque benchmark.
| Type | Claude Sonnet 4.5 | GPT-5 Nano |
|---|---|---|
| Provider |  Anthropic |  OpenAI |
| score moyen | 42.1 | 45.3 |
| Prix d'entrée | $3.00 | $0.05 |
| Prix de sortie | $15.00 | $0.40 |
| Fenêtre de contexte | 1.0M tokens (~500 books) | 400K tokens (~200 books) |
| Sorti le | 2025-09-29 | 2025-08-07 |
| Code source ouvert | Proprietary | Proprietary |
Scores de benchmark
12 benchmarks · Claude Sonnet 4.5: 11, GPT-5 Nano: 1
| Benchmark | Catégorie | Claude Sonnet 4.5 | GPT-5 Nano |
|---|---|---|---|
| ARC-AGI | reasoning | 63.7 | 20.7 |
| ARC-AGI-2 | reasoning | 13.6 | 2.6 |
| FrontierMath-2025-02-28-Private | math | 15.2 | 8.3 |
| FrontierMath-Tier-4-2025-07-01-Private | math | 4.2 | 2.1 |
| GPQA diamond | knowledge | 76.4 | 59.3 |
| MATH level 5 | math | 97.7 | 95.2 |
| OTIS Mock AIME 2024-2025 | math | 77.8 | 81.1 |
| SimpleQA Verified | knowledge | 23.6 | 12.2 |
| SWE-Bench Verified (Bash Only) | coding | 70.6 | 34.8 |
| Terminal Bench | coding | 46.5 | 11.5 |
| VPCT | knowledge | 9.7 | 5.8 |
| WeirdML | coding | 47.7 | 38.1 |